摘要
由于缺少以较少的存储空间描述汉字特征概率密度函数方法,在汉字识别系统中贝叶斯分类器很少得到应用.本文提出一种只需6个数据就能描述一个特征的概率密度函数的方法,构造了一个基于贝叶斯分类器的手写汉字识别系统.实验结果表明,该分类器具有优良的性能.
Since lack of effective methods to describe the probability density functions of the features of Chinese characters, Bayesian classifiers are rarely used in the Chinese character recognition systems. In this paper, a method, which can describe the probability density function of a feature with only 6 digits, is proposed. Based on this method, a handwritten Chinese character recognition system using a Bayesian classifier is constructed.In our experiments, the classifier showed good performances.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2002年第12期1804-1807,共4页
Acta Electronica Sinica
基金
国家863计划(No.2001AA114080)
教育部科学技术研究重点项目(No.02029)